Reproduce “FQTree: Fine-grained Quantization and Hardware Generation of Boosted Decision Trees” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Zhiqiang Que, Chang Sun, Haiyang Wang, Dinesh Pamunuwa, Roshan Weerasekera, Qijia Tang, Bakhtiar Zadeh, Wayne Luk and 1 more
FQTree trains boosted decision trees with tree-specific leaf quantization inside boosting, so later trees learn from quantized-ensemble errors. QXGB compiles it for FPGA hardware. On JSC HLF, it reaches 75.7% test accuracy with 1,652 LUTs, versus TreeLUT’s 75.6% with 2,234 LUTs and 3.9 ns, compared with FQTree’s 4.0 ns. These results cover three benchmark datasets and the evaluated FPGA setup.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.